The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains deceptive metadata in SKILL.md claiming to be an "official" resource from the expo/skills repository and using the @expo NPM namespace. Since the skill is authored by sickn33, this impersonation is a serious trust violation that could lead users or agents to grant unnecessary permissions or trust to the skill.
[COMMAND_EXECUTION]: The instructions require the agent to execute shell commands to install dependencies (npm install) and run local utility scripts (fetch.js and validate.js). This requirement for local code execution increases the attack surface of the execution environment.
[EXTERNAL_DOWNLOADS]: The skill retrieves technical documentation and JSON schemas from official Expo endpoints (api.expo.dev and github.com/expo/expo). These are recognized well-known sources for the Expo development platform and are used to provide the validation logic for the skill.
[DATA_EXFILTRATION]: The fetch.js utility allows for arbitrary URL fetching and local caching. This tool could be leveraged as a data exfiltration vector if the agent is manipulated into appending sensitive data to URL parameters when calling the script.
[PROMPT_INJECTION]: The skill ingests and validates untrusted user-provided YAML workflow files. This processing creates an indirect injection surface where malicious instructions embedded in a workflow file could attempt to influence the agent's behavior during the validation or explanation process.

expo-cicd-workflows